Evaluating macroinvertebrate biological metrics for ecological assessment of streams in northern Portugal

DOI 10.1007/s10661-009-0996-4

Evaluating macroinvertebrate biological metrics

for ecological assessment of streams in northern Portugal

Received: 26 November 2008 / Accepted: 13 May 2009 / Published online: 2 June 2009 © Springer Science + Business Media B.V. 2009

Abstract A procedure to select the most relevant metrics for assessing the ecological condition of the Douro basin (north Portugal) was developed based upon a set of 184 benthic community met-rics. They were grouped into 16 biological cate-gories selected from literature using data collected over 2 years from 54 sites along 31 rivers covering the whole perceived range of human disturbance. Multivariate analyses were carried out to identify the main trends in the macroinvertebrate data, to select reference versus impaired sites, to avoid multicolinearity between metrics, and to identify those that were clearly independent from natural stream typology. Structural metrics, adaptation metrics, and tolerance measures most effectively responded across a range of human influence. We find these attributes to be ecologically sound for monitoring Portugal’s lotic ecosystems and pro-viding information relevant to the Water

Frame-Electronic supplementary material The online version

of this article (doi:10.1007/s10661-009-0996-4) contains supplementary material, which is available to

authorized users.

S. G. Varandas (

_B

CITAB—Center for Research and Technology of Agro-Environment and Biological Sciences, Forestry Department, Universidade de Trás-os-Montes e Alto Douro, Apartado 1013, 5001-801, Vila Real, Portugal

e-mail: simonev@utad.pt

work Directive, which asserts that the definition of water quality depends on its “ecological status”, independent of the actual or potential uses of those waters.

Keywords Attributes· Traits · Benthic fauna · Ecological condition· Biomonitoring ·

Aquatic ecosystems

Introduction

Over the past century, increasing stress on ecosys-tems has resulted in persistent efforts to find a universal indicator capable of evaluating river quality to use as a tool able to detect impacts on biological resources. Important biological com-ponents have been measured and several indices have been developed in the light of these studies. However, none of the attributes used up until now contain a sufficient array of responses capable of distinguishing between different types of impacts (Dolédec et al. 1999). Freshwater biomonitoring has focused on changes in community structure and composition (distribution and abundance). Multimetric approaches integrate several descrip-tors of the sampled assemblage (Karr1991; Niemi and McDonald 2004), which can synthesize and interpret large amounts of information and cover effects of multiple stressors (Bonada et al. 2006,

et al. (1995) and Davis et al. (1996) have also verified that, for a broad range of human impacts on aquatic ecosystems, combined multiple met-rics provide an effective way of assessing water resources. Multimetric techniques have been crit-icized due to the loss of ecological information caused by aggregating metrics into a unique index value (Norris1995; Suter1993). Other techniques that rely on multivariate statistical analyses have also been worldwide developed (Wright 1995; Cortes et al. 1998; Oliveira and Cortes 2006; Peeters2001). Multivariate approaches are based upon statistical relationships between fauna and selected environmental characteristics to discern patterns in community composition. Furthermore, the multivariate analyses provide a statistical objective method for grouping sites with simi-lar macroinvertebrates communities (Reynoldson et al.1997). Examples of complex but intensively used systems include the River Invertebrate Pre-diction and Classification System (Wright 1995) used in UK and its derivative, the Australian River Assessment Scheme models (Simpson and Norris 2000) used in Australia and Benthic As-sessment of Sediment (Reynoldson et al. 1995) used in parts of Canada.

In recent years, there has been a return to basic ecology, using species traits and functional groups in ecological studies (e.g., Lenat 1993; Palmer et al. 1996; Poff and Allan 1995; Statzner et al.

1994, 1997; Pont et al. 2006; Tomanova et al.

2006). Besides recording species loss or reduction as a result of stress conditions, species traits can pinpoint life history characteristics, more sensi-tive to disturbance (Phillips2004). As Poff (1997) states, “traits presumably represent functional re-lationships with important environmental selec-tive forces, such as stream flooding or drying, local shear stress, temperature extremes, and human pollution”. Reinforcing this idea, Bonada et al. (2007) found that studies at larger spatial scales and higher biological organizational levels using biological traits can anticipate impacts of climate change.

Based upon these findings, a logical step for-ward would be to test, for the purpose of bioassessment alongside conventional metrics, new attributes (traits of reproduction, life cycle, growth, locomotion, dispersion, etc.) that

repre-sent adaptive responses to environmental factors. This study aims to exploit the strengths of both multimetric and multivariate procedures for as-sessing the suitability of metrics in evaluating the ecological status of water courses. This inte-grated approach has been used by authors such as Zamora-Munoz and Alba-Tercedor (1996) and Cortes et al. (2002b) namely a multivariate ap-proach to classify sites a priori and establish refer-ence conditions based on environmental variables followed by a multimetric method, based upon a wide range of community characteristics, to then select core metrics by comparing community structure at reference sites and impaired sites.

A nationwide surface water quality monitor-ing is coordinated by the National Institute of Water (INAG). Until the implementation of Water Framework Directive (WFD; Directive 2000/60/CE—European Commission 2000), the national network consisted of 109 sampling sites, primarily in large rivers (24 river systems). The most important reservoirs whose main use is do-mestic water supply are included in the program as well. Water samples were taken at monthly intervals and analyzed only for general chemical and physical variables, organic pollution indica-tors, nutrients, and some heavy metals. The bio-monitoring programs that have been carried out almost always are restricted to biological indexes developed for other countries such as Iberian Biological Monitoring Working Party (BMWP’; Alba-Tercedor 2000) and Belgian Biotic Index (BBI; De Pauw and Vanhooren1983).

Nowadays, under WFD, Portugal is required to develop programs to evaluate physical, chem-ical, and biological integrity and to adopt water quality standards to restore and maintain that integrity. The WFD is the most important new European legislation concerning aquatic resource management to emerge for decades. The aim of the WFD is long-term sustainable water manage-ment based on a high level of protection of the aquatic environment (inland and coastal waters) and prevents further deterioration through better land management. Successful implementation of the WFD will go a long way to protecting and enhancing the quality of all stages of the water cycle in a sustainable manner. Current European Water Policy addresses the increasing awareness

and participation of citizens, stakeholders, and other involved parties in decision making. The WFD is innovative since it asserts that the defi-nition of water quality depends on its “ecological status” independent of the actual or potential uses of those waters. Ecological status derives from the assessment of surface water bodies, defined by the global expression of the structure and function of selected biological communities, taking into account geographical and climatic factors as well as the physical and chemical descriptors, including those resulting from human activities. This ap-proach takes into consideration the overlapping unequal and dynamic influence of the three key components (physical, chemical, and biological integrity), similar to the pattern established after Yoder (1995; from Barbour and Yoder2000). The proposed methodology aims to provide tools that

contribute to the implementation of the WFD in defining monitoring programs.

Materials and methods

Study sites and sample collection

Data were collected from the Portuguese section of the Douro catchment and covered all the ma-jor tributaries (Fig. 1). The main river of this catchment (river Douro) flows generally westward across Spain and northern Portugal to the Atlantic Ocean at Foz do Douro. It has extensive large traffic in its Portuguese section and is fully reg-ulated for hydropower purposes. Tributaries are small and flow into canyons to enter the larger river. The main impacts of the study area are the

Fig. 1 Location of the Douro basin in north Portugal, showing sampled sites. Reference (star) and highly impaired sites

strong urban pressure in the coastal strip creat-ing environmental problems with very negative effects in the running water. The remaining area is dominated by agroforestry and low population

density. The main characteristics of reference and impacted sites are summarized in Table1.

Benthic macroinvertebrates were sampled in summer 1997 and 2000 at 54 sites from 31 rivers

Table 1 Main characteristics of studied sites of the Douro basin (reference and impacted sites)

Method Reference sites Impaired sites

mean (min–max) mean (min–max)

Altitude (m) Measured from 1:25000 535 (121–823) 310 (76–653)

topographic maps

Mean catchment slope Van Haveren (1986) method 0.42 (0.20–0.74) 0.52 (0.28–0.85)

Stream order Strahler system measured in 3 (2–4) 4 (3–6)

1:250,000 topographic maps

Mean wetted channel Distance from bank to bank 6.6 (2.6–14.6) 20.2 (6.8–80) width (m) at a transect representative

of the stream width in the sampled area using a tape measure

Mean water depth (m) Measured at 3 transects over 0.45 (0.17–0.60) 0.74 (0.35–1.45) the sampled reach using

a graduated stick

Distance from Distance from source (headwater) 28 (10–63) 71 (40–110) source (km) to the sampling site

measured along the river from 1:250,000 topographic maps

Mean daily air Average air temperature 12.3 (10–15) 14.2 (12–16) temperature (◦C) measured during the period

1931–1960 from 1:1,000,000 Atlas do Ambiente. It was used the maximum value of each class

Mean annual Average annual precipitation 1035 (650–1300) 745 (550–1400) precipitation (mm) measured during the period

1931–1960 from 1:1,000,000 Atlas do Ambiente. It was used class midpoint value

Anthropogenic Each perturbation factor in the 2.7 (0–8) 5.9 (0–15) perturbation—anthP neighboring area to the reach

(e.g., agriculture, urban zone, road beds, pasture, inert extraction, parking lots, deforestation, reforestation, waste, clear cutting,...) received the weight of 1 or 2 if it was present less or more than 10%, respectively. The final score was obtained by the sum of all values of the both slopes

Riparian canopy Percentage of wetted bed shaded 68 (36–96) 8 (0–45) cover—%ripC by riparian vegetation, measured

Table 1 (continued)

Method Reference sites Impaired sites

mean (min–max) mean (min–max) SAV It was used 6 categories (0= no 20.8 (10–25) 11.2 (1.5–25)

trees;...; 5= continuous arboreal vegetation) weighted by scores: 5 (present in more than 33% of the area) or 3 (less than 33% of the area is occupied by that category). Evaluation of both banks. The final score was obtained using the following formula: SAV= lbi×pj n + rbi×pj n 2 where

n= number of the observed

classes for each bank lbi= class in the left ban

rbi= class in the right bank pj= 4 weight

QBR Index developed by 85 (60–100) 60 (20–90)

Munné et al. (2003)

UA Impact of urbanization on river. 1 (1–2) 2 (1–4)

Classes 1 to 5: 1= ≤1% urban land; 2= low impact (≤10% urban land); 3= moderate impact (10–20% urban land); 4= strong impact (20–40% urban land); 5= Severe impact (≥40% urban land)

SLS Deviation from natural stream 1 (1–2) 2 (2–3)

bottom as a result of deposition. Classes 1 to 5 with a score of 5 representing maximum deviation and 1 the minimum alteration

Affluent load generated BOD—amount of biochemical 45.4 (7.5–139.3) 750.7 (86.7–1624) by human population degradable substances (chemical 64.8 (10.7–198.7) 1070.7 (124–2315) (t year−1) stressor) that flow into the stream; 9.1 (1.5–27.8) 149.8 (27.3–324.1) BOD5 TSS—amount of sediments that 2.8 (0.5–8.5) 45.8 (5.3–98.9) TSS come into the stream and remain

Ntot in suspension in water column.

Ptot Ntotand Ptot—amount of total nitrogen and total phosphorous that comes into the stream, respectively. Firstly was calculated the generated load: total number of inhabitants for drainage basin multiplied by capitation—g/inhab day (coefficient used in the Water National Plan, 2001). Secondly, the resulting value was multiplied by 1 minus the proportion of removed domestic load by treatment stations of residual waters (the appropriate conversion coefficients were used)

distributed along the whole range of perceived en-vironmental conditions. To assess anthropogenic impacts, attributes describing riparian vegeta-tion, water quality, and instream variables were recorded at each site (some parameters are de-scribed in Tables1and5). Composite multihabitat samples were taken at each site using a hand net (350 μm mesh size). Each habitat (e.g., rif-fle, pool, edge, vegetation) was sampled in pro-portion to its representation. Hand-net contents were store in sample containers and refrigerated, once in the laboratory macroinvertebrates were sorted live and preserved. All organisms were removed, counted, and identified using standard keys developed for the Iberian Peninsula, usu-ally to species level of taxonomic resolution, with the exceptions of Hydracarina (order), Diptera (family, subfamily, or tribe for Chironomidae), Oligochaeta (family), and Coleoptera (genus or species).

Identification of potential benthic metrics

A total of 184 macroinvertebrate assemblage attributes were selected for testing as poten-tial metrics. These attributes included 16 bio-logical categories (“Electronic supplementary material”) ranging from those traditionally tested and integrated in multimetric indices (richness, composition and dominance measures, tolerance and intolerance descriptors, and biotic indices) to species traits. A broader set of inverte-brate traits were studied (respiration, resistance forms, locomotion, and substrate relation—habit/ behavior, habitat preference, forms to avoid the drift, functional feeding groups, life cycle characteristics, aquatic stages, reproduction met-rics, dispersal measures, and maximal size) since that can be applied to larger temporal and spatial scales which may vary widely between ecoregions (Charvet et al. 2000; Statzner et al.

2005; Bonada et al. 2006). Selected metrics

were compiled from a wide variety of works including Kerans and Karr (1994), Resh (1994), Statzner et al. (1994), Barbour et al. (1995,1996, 1999), Fore et al. (1996), and Kashian and Burton (2000). Benthic macroinvertebrate traits were based on Charvet et al. (2000), Dolédec et al. (1999), and Usseglio-Polatera et al. (2000).

Functional feeding groups and habit/behavior were assigned according to the primary cate-gory documented by Tachet et al. (2002) and data bases supplied by “Usseglio-Polatera in 2005” and complemented with information based upon Merritt and Cummins (1996) and Barbour et al. (1999). Categorization according to sensi-tivity to pollution was based upon Alba-Tercedor and Sanchez-Ortega (1988), Cortes (1992), and Monzón (1996). Three biological indices were ap-plied: the BMWP’, BBI, and Family-Level Biotic Index (FBI; adapted from Hilsenhoff1988). Fam-ily tolerance scores for the FBI index were used in agreement with Alba-Tercedor (2000), since they are adapted for the Iberian fauna. Reproduction metrics, resistance forms, mechanisms to avoid drift, dispersal measures, and maximal size were divided into classes (see “Electronic supplemen-tary material”) and calculated based on informa-tion described mainly in the last update of data bases supplied by “Usseglio-Polatera in 2005”. Respiratory physiology and habitat preference macroinvertebrate classifications were based upon Hynes (1979), Richoux (1982), Margalef (1983), Faessel (1985), Askew (1988), Chinery (1992), Wetzel (1993), Fitter and Manuel (1994), Nieser et al. (1994), and Vieira-Lanero (2000). Other publications such as Thorp and Covich (1991) and online papers found via online literature searches were used to complement existing data or provide missing data. When conflicting natural history in-formation for the same taxon occurred, we con-sidered both sources of information. So, the two more important designations (except for life cycle duration that was considered the maximum of three designations) were used. Data were divided by the total number of designations.

All the above information used for assemblage characterization was supplemented and in some cases modified by specific information relating to local fauna and information found in literature and online sites. When ambiguities concerning a particular attribute could not be resolved, data were discarded. Whenever possible, metrics were defined at species level. Some of them included higher taxonomic levels (see “Electronic supple-mentary material”). Metrics were expressed in quantitative terms, most of them either as relative proportions or number of taxa.

Data analysis

Identification of reference and impaired sites Definition of human disturbance categories

Dis-turbed and reference (or minimally disDis-turbed) sites were identified, firstly, based on a crite-rion to assess the human impact of European rivers (FAME Consortium 2004) linked to the implementation of the WFD and in the princi-ples of REFCOND (2003) and, secondly, by their physicochemical characteristics, for a consistent and independent split. By using the first crite-rion, sites were classified according to anthro-pogenic disturbance by quantifying ten stressors that covered from local impacts (observed in situ) to the ones at the catchment level determined from geographical information system, to assess impacts in the river basin upstream. The list of descriptors included land use intensification, ur-ban area, structure of riparian layer (including invasive plants), river connectivity, sediment load, hydrological modifications (water abstraction and flow regulation), symptoms of acidification or tox-icity, morphological condition, symptoms of eu-trophication, and invasive plant or animal species. Each variable was allocated to one of five classes according to the magnitude of the stressor under evaluation (1 corresponds to high status= refer-ence conditions: only minor, negligible alterations, and 5 indicates a bad status: severe impact). This procedure allowed to define the reference sites (a) according to their total score, derived from rank-ing the scores of environmental degradation, and (b) when none of the above variables occurred in the two “worst” classes.

Finally, collected physical and chemical data were compared with the national Water Institute (INAG) water quality classification scheme which uses 27 parameters that describe principal nutri-ents and micropollutants, to classify water quality according to uses. Thus, a site was considered unimpaired if it belonged to class A, impaired if it belonged to classes B or C and highly impaired if it belonged to classes D or E.

Species ordination The relation of the

macroin-vertebrate communities with the previous

classi-fication of sites produced by the environmental variables was assessed by multivariate ordina-tions of species of taxa abundance structure (log (x+ 1) transformed) for each year, using non-metric multidimensional scaling analysis (NMDS) based on similarity scores (Bray–Curtis coeffi-cient) using PRIMER v 5.2.2 (Clarke and Gorley 2001). Separation of reference and the most de-graded sites was further refined by integrating the 1997 and 2000 data and checking for interan-nual variability. Reference sites with interaninteran-nual consistency along disturbance categories were re-tained. A similarity percentages (SIMPER) analy-ses (SIMPER-species contributions) was done to quantify the degree of differences found between years or quality categories, as well as species that contribute more to these differences. This analy-sis although not perform formal test of hypothe-ses provides a list of species in order of their percentage contribution to dissimilarities (Bray– Curtis dissimilarity) between groups or similar-ities within groups (Clarke and Warwick 2001). Metrics-derived values were then compared be-tween reference and impaired sites defined above.

Evaluation of metrics

Biological patterns and human disturbance The

optimal combination of potential metrics for dis-criminating between reference and degraded sites was determined using detrended correspondence analysis (DCA; Hill and Gauch1980). This analy-sis evaluates the ecological sensitivity and re-sponse direction of all potential benthic attributes to increasing human stress for each year, extract-ing the most relevant biological patterns, and then relating them to patterns induced by human activ-ity. DCA provides a more precise representation of environmental gradients than principal compo-nent analysis, since it arranges the data in such a way to avoid the compression of the ordination axes and quadratic distortion (Hill1979). The data were log (x+ 1) transformed with the exception of metrics expressed in percentage. These metrics were not transformed since they were normally distributed (Shapiro–Wilk test for normality).

Assessment of metrics Pearson correlation

disturbance gradient and test metrics were de-rived; only significant correlations ( p≤ 0.05) were retained. The positive or negative signs of the Pearson correlation coefficients indicated the di-rection of the candidate metric (i.e., increasing or decreasing in response to perturbation). When signals of the two coefficients (year 1997 and year 2000) were not coincident, the response was considered as variable and eliminated, unless we knew the cause of variation. The DCA was com-puted using the package CANOCO version 4.0 (Ter Braak and Smilauer1998). Metrics responses that were not in accordance with literature were also discarded.

Candidate metrics were further tested for eco-logical consistency. First, a coefficient of variation (CV) was calculated for each metric in the group of all reference sites that presented interannual consistency, as a measure of within-site variability. CVs were calculated from the multiyear reference data, once this combination had absorbed the interannual variability verified previously. Low CV is an important condition for determining the suitability of a metric in detecting anthropogenic impacts. Kashian and Burton (2000) state that CVs > 50% are ineffective in detecting between impaired and unimpaired conditions. Based on this criterion, only candidate metrics with CVs

< 50% were retained for further analyses.

Based on the approach advocated by Barbour et al. (1996), metrics with many zero or low values in the reference sites were assessed to avoid nondetection of lower values. Metrics that followed this pattern were also eliminated from the subsequent tests (“Electronic supplementary material”).

Obtaining a subset of independent metrics A

preliminary canonical correspondence analysis (CCA) with forward selection (stepwise proce-dure) was performed, for both datasets (metrics and taxa abundance) containing the 2 years of samplings (1997 and 2000 combined), to elimi-nate metrics that did not make any significant contribution to explaining biological variation derived from disturbance. Only metrics with p≤ 0.05 were selected for the next step. In order to distinguish between natural typological variability from perturbation-induced changes, typological

variables such as distance from source, stream width, altitude, stream order, zonation (accord-ing to Illies and Botosaneanu 1963), catchment area upstream of the site, and total stream length were considered as covariables in this CCA. The resulting variation inflation factors (VIFs) were assessed. Metrics with high VIFs (>20) values imply redundancy (multicolinearity) with other metrics, a phenomenon that should be avoided since no unique contribution is made to the regression equation (Ter Braak and Smilauer 1998). Consequently, the metric with the high-est VIF was excluded and the CCA procedure was repeated until all metrics exhibiting multico-linearity were excluded. The environmental and biological variables (metrics) were previously standardized [(x − ¯x) ÷ sd (x)] to achieve compa-rable scales and the macroinvertebrates abun-dance were transformed by log (x+ 1). The package CANOCO version 4.0 was used for the CCA analyses.

Establishing the relationship between metrics and human disturbance gradient Following the

met-rics selection procedure using CCA, Pearson correlations were derived between variables de-scribing human disturbance and the preselected metrics, ensuring that only metrics that were good indicators of human impacts were retained. Variables used to test metrics were conductivity (cond), riparian canopy cover (%ripC), anthro-pogenic perturbation (anthP), structure of arbo-real vegetation (SAV), riparian habitat quality (QBR index; Munné et al. 2003), urbanization area (UA; categories 1–5), sediment load segment (SLS; categories 1–5), dissolved oxygen (O2), tur-bidity, upstream dams (UD; categories 1–5), and effluent load generated by human population: biochemical oxygen demand (BOD5), total sus-pended solids (TSS), total nitrogen (Ntot), and total phosphorous (Ptot). These variables and an-other ones (water temperature, substratum size, mean water depth and velocity, stream width, and stream flow) were measured simultaneously with invertebrates sampling. Finally, benthic attributes that best discriminated between the reference and degraded sites were illustrated by box-and-whisker plots. This method, described by Barbour

et al. (1996), rejects metrics if the interquartile range of impaired or reference sites overlaps with the median of the other. Results from the 2 years were compared and only metrics that detected a gradient of human impact in both years were accepted as core metrics with potential to be part of a biological indicator system.

Results

Selection of reference and impaired sites

NMDS based on species/site data of 1997 (2D stress value = 0.23, 3D stress value 0.17) and

Highly Impaired sites Reference sites b) Rabaçal 1 Rabaçal 2 Reference sites Highly Impaired sites a)

Fig. 2 NMDS ordination plot of sites a sampled in 1997.

Outliers Rabaçal 1 and Rabaçal 2 were excluded from the impaired and reference sites, respectively, because they had low abundance and diversity probably caused by sam-pling difficulties. b Sampled in 2000. The curved lines rep-resent an imaginary separation between the reference and impaired sites from the remaining ones. Identical symbols for reference and impaired sites were used to illustrate the same places

Impaired Reference 1997

2000

Fig. 3 NMDS ordination plot of multiyear data

represent-ing the variability between years and the separation be-tween reference and impaired sites. The dot line represents an imaginary separation among years 1997/2000

data from 2000 (2D stress value= 0.19, 3D stress value= 0.13) differentiated between undisturbed and highly impaired sites (Fig.2a, b). That is, iden-tified sites in the reference and highly impaired groups (ten sites in each category) for each year were basically the same (Fig. 2). The distinction between these two subsets of sites was confirmed with SIMPER tests performed in the multiyear NMDS ordination. NMDS analysis displayed a gradient of disturbance with the sites distributed along a gradient of human impacts of increasing magnitude. Average dissimilarity between 1997 and 2000 was 77.09%, making clear the existence of a strong interannual variability (Fig.3; Table2). A total of 174 species (43% of the total taxa)

Table 2 Percentage breakdown of average dissimilarity

between groups of years (1997 against 2000) and groups of impaired versus reference sites in Douro catchment, using SIMPER analysis

Factors Groups Average Average similarity (%) dissimilarity (%)

Years 1997 25.07 77.09

2000 33.30

Ref/Imp Reference 34.24 79.89 Impaired 29.43

Table 3 Average contribution of species principally responsible for intragroup similarities within each year of sampling

1997 2000

Species Contribution (%) Species Contribution (%)

Chironomini gen. sp. 14.81 Chironomini gen. sp. 11.11

Hydracarina 11.55 Caenis luctuosa 7.99

Tanytarsini gen. sp. 7.46 Tanytarsini gen. sp. 7.91

Tanypodinae gen. sp. 6.96 Tanypodinae gen. sp. 6.59

Orthocladiinae gen. sp. 5.33 Hydracarina 5.49

Caenis luctuosa 4.49 Orthocladiinae gen. sp. 5.44

Lumbriculidae gen. sp. 3.50 Diptera pupae 3.43

Dryops sp. 2.71 Aquarius najas 3.32

Physella acuta 2.17 Hydropsyche sp. 2 2.60

Mystacides azurea 1.97 Simuliidae gen. sp. 2.42

Calamoceras marsupus 1.80 Lumbriculidae gen. sp. 2.15

Oulimnius sp. 1.79 Baetis fuscatus 1.77

Laccophilus sp. 1.67 Mystacides azurea 1.72

Diptera pupae 1.66 Gerridae juvenil 1.64

Gomphus pulchelus 1.59 Ecdyonurus gr. venosus 1.53

Lumbricidae gen. sp. 1.54 Leuctra aurita 1.41

Aquarius najas 1.32 Choroterpes picteti 1.32

Gerris sp. 1 1.23 Ancylus fluviatilis 1.23

Tubificidae gen. sp. 1.18 Laccophilus sp. 1.22

Hydropsyche sp. 2 1.11 Cloeon gr.simile 1.21

Atyaephyra desmaresti 1.09 Baetis rhodani 1.13

Platycnemis latipes/acutipennis 1.07 Hydroptilidae gen. sp. 1.12

Leuctra fusca 1.02 Oulimnius sp. 1.10

Boyeria irene 0.95 Cloeon gr. dipterum 1.01

Euleuctra geniculata 0.78 Ephemerella ignita 0.94

Atrichops sp. 0.77 Baetis sp. 0.88

Baetis rhodani 0.69 Habrophlebia eldae 0.86

Onychogomphus uncatus 0.68 Boyeria irene 0.81

Erpobdella monostriata 0.60 Rhyacophila sp. 0.77

Coenagrion puella 0.57 Polycentropus flavomaculatus 0.75

Calopteryx virgo 0.54 Naucoris maculatus maculatus 0.73

Stenelmis canaliculata 0.53 Leuctra geniculata 0.72

Ecdyonurus gr.venosus 0.49 Hydrometra stagnorum 0.72

Corixidae gen. sp. 0.49 Enchytraeidae gen. sp. 0.72

Atherix sp. 0.47 Erpobdella monostriata 0.64

Anacaena sp. 0.46 Gerris lateralis 0.61

Simuliidae gen. sp. 0.42 Physa fontinalis 0.61

Leuctra sp. 1 0.42 Micronecta scholtzi 0.55

Haliplus sp. 0.42 Calamoceras marsupus 0.52

Naucoris maculatus maculatus 0.42 Platycnemis latipes/acutipennis 0.50

Hydrochus sp. 0.40 Onycogomphus uncatus 0.47

Nepa cinerea 0.39 Tubificidae gen. sp. 0.46

Platycnemis cf. latipes 0.35 Pseudocentroptilum pennulatum 0.42

Potamopyrgus jenkinsi 0.34 Leuctra sp. 1 0.41

Epeorus silvicola 0.39

Dryops sp. 0.37

Allogamus ligonifer 0.31

Table 4 Average contribution of species principally responsible for intragroup similarities within reference and highly

impaired sites

Reference Impaired

Species Contribution Abundance Species Contribution Abundance

(%) class (%) class

Chironomini gen. sp. 6.66 − Chironomini gen. sp. 18.94 +

Hydracarina 6.59 − Caenis luctuosa 15.76 +

Tanytarsini gen. sp. 5.63 − Hydracarina 12.38 +

Orthocladiinae gen. sp. 4.93 − Tanypodinae gen. sp. 6.68 +

Tanypodinae gen. sp. 4.37 − Orthocladiinae gen. sp. 5.87 +

Calamoceras marsupus 3.77 + Tanytarsini gen. sp. 5.84 +

Oulimnius sp. 2.99 + Atyaephyra desmaresti 4.46 +

Habrophlebia eldae 2.88 + Diptera pupae 2.98 +

Aquarius najas 2.81 + Lumbriculidae gen. sp. 2.91 +

Leuctra sp. 1 2.78 + Physella acuta 1.69 +

Ecdyonurus gr. venosus 2.48 + Aquarius najas 1.45 −

Polycentropus sp. 1 2.45 + Hydropsyche sp. 3 1.36 +

Dryops sp. 2.36 − Coenagrion puella 1.34 +

Onycogomphus uncatus 2.36 + Mystacides azurea 1.17 +

Atherix sp. 2.29 + Choroterpes picteti 1.04 +

Hydropsyche sp. 2 2.25 + Micronecta scholtzi 0.96 +

Leuctra geniculata 2.18 + Ecdyonurus gr. venosus 0.87 −

Baetis rhodani 2.08 − Platycnemis latipes/acutipennis 0.81 +

Boyeria irene 1.98 + Gomphus pulchelus 0.81 +

Caenis luctuosa 1.84 − Simuliidae gen. sp. 0.77 +

Hydraena sp. 1.83 − Laccophilus sp. 0.76 +

Leuctra aurita 1.70 + Gerris sp. 0.74 +

Simuliidae gen. sp. 1.52 − Cloeon gr. simile 0.67 +

Sericostoma sp. 1 1.50 + Rhyacophila sp. 1 1.43 + Diptera pupae 1.42 − Elmis sp. 1.32 − Orectochilus villosus 1.18 − Ancylus fluviatilis 1.17 − Mystacides azurea 1.13 − Ephemerella ignita 1.12 − Polycentropus flavomaculatus 1.00 − Allogamus ligonifer 0.99 + Lumbriculidae gen. sp. 0.93 − Epeorus silvicola 0.80 + Dixidae gen. sp. 0.72 − Baetis sp. 2 0.63 + Aphelocheirus occidentalis 0.63 + Calopteryx virgo 0.59 + Limnius sp. 0.58 + Leuctra fusca 0.54 + Protonemura meyeri 0.52 + Gerridae gen. sp. 0.43 − Baetis fuscatus 0.40 − Dupophilus brevis 0.40 +

The signals+ (increase) or − (decrease) in column “Abundance class” indicate which species are more or less abundant in reference or impaired sites

accounted for 90% of the dissimilarity between reference and impaired sites, and 74 (18% of the total of species) contributed for 66% of the dissimilarity. SIMPER results of reference versus impaired sites (Fig. 3; Table 2) correspond with the pattern observed in the previous analyses. A value of 79.89% dissimilarity between the most and the least disturbed sites confirms that these reference and impaired sites groups are truly dis-tinct despite strong interannual variation. These observations correspond with our initial findings, based on groups established using physicochemi-cal data.

Principal representative macroinvertebrate taxa of the reference and impaired groups (to a cumulative percentage of 90%) are presented in Tables 3 and 4. Reference sites are typified by intolerant species belonging to Trichoptera, Ephemeroptera, Plecoptera, Coleoptera, and Heteroptera such as Calamoceras marsupus (Brauer), Sericostoma sp. (Latreille), Allogamus

ligonifer (McLachlan), Habrophlebia eldae (Jacob

and Sarton), Ecdyonurus gr. venosus (Fabricius),

Leuctra aurita (Navas), Protonemura meyeri

(Pictet), Oulimnius sp. (Gozis), Elmis sp. (Latreille), Limnius sp. (Illiger), Dupophilus

brevis (Mulsant), Aquarius najas (Degeer), and Aphelocheirus occidentalis (Nieser and Millán;

Table 4). Disturbed sites are characterized almost exclusively by tolerant species from across several orders such as Diptera, Ephemeroptera, Trichoptera, and Heteroptera Table 4. On the other hand, the species that most contributed to dissimilarities between reference and impaired sites were associated to intolerant species (e.g.,

Leuctra sp., Sericostoma sp., C. marsupus, A. najas, Brachycentrus subnubilus). Only 15% of

the total of the species data (65 species) had contributed with 2/3 to the dissimilarities among sites.

Metric selection

A consistent pattern of disturbance in the ordi-nation space was displayed by DCA analyses for each year, i.e., sites were distributed in the same order along the disturbance gradient. The Pearson coefficients of first axes correlations with all po-tential untransformed metrics showed a positive Table

5 Pearson correlations between selected metrics (indicated o n “ Electronic supplementary material ”) through CCA analyses and variables of human pressure fP fH fEPT %5TD %br % diap fSwi fCling % Cling % exR %preR g Shr sfil fGath %Int IBMWP FBI % Hem % CluV %aqP cond − 0. 43 − 0. 12 − 0. 48 0.30 0.01 − 0. 08 − 0. 09 − 0. 26 0.09 − 0. 30 − 0. 20 − 0. 46 − 0. 31 − 0. 38 − 0. 44 − 0. 44 − 0. 49 − 0. 26 − 0. 31 0.20 (μ Sc m − 1) %ripC 0.39 0.10 0.41 − 0. 15 − 0. 11 − 0. 17 0.31 0.37 0.22 0.34 0.02 0.52 0.25 0.38 0.53 0.49 0.38 − 0. 20 0.20 0.10 anthP − 0. 25 − 0. 11 − 0. 35 0.39 − 0. 17 − 0. 12 − 0. 26 − 0. 32 − 0. 01 − 0. 28 − 0. 14 − 0. 21 − 0. 31 − 0. 13 − 0. 42 − 0. 30 − 0. 40 − 0. 33 − 0. 37 0.07 SAV 0 .26 0 .07 0 .29 − 0. 14 − 0. 21 0.05 0.09 0.17 0.03 0.24 − 0. 01 0.35 0.09 0.25 0.29 0.34 0.17 − 0. 23 0.27 0.14 QBR 0 .48 0 .10 0 .48 − 0. 32 0.03 0.01 0.16 0.43 0.07 0.40 0.18 0.37 0.35 0.14 0.55 0.48 0.45 0.17 0.31 0.02 UA 0.05 − 0. 12 − 0. 10 0.27 − 0. 38 − 0. 09 − 0. 18 − 0. 06 − 0. 01 − 0. 03 0.03 − 0. 02 − 0. 06 0.01 − 0. 18 − 0. 03 − 0. 20 − 0. 32 − 0. 05 0.29 SLS − 0. 50 − 0. 02 − 0. 53 0.34 − 0. 13 0.14 − 0. 08 − 0. 52 − 0. 07 − 0. 43 − 0. 13 − 0. 27 − 0. 45 − 0. 26 − 0. 48 − 0. 51 − 0. 44 − 0. 22 − 0. 29 0.07 O2 (mg/L) − 0. 08 − 0. 13 − 0. 10 0.14 0.03 0.00 − 0. 21 − 0. 13 0.19 0.04 0.06 − 0. 27 0.10 − 0. 24 − 0. 05 − 0. 24 − 0. 10 − 0. 11 − 0. 24 0.23 Turbidity 0 .56 − 0. 02 0.58 − 0. 15 − 0. 10 − 0. 18 − 0. 02 0.38 0.08 0.30 0.29 0.53 0.35 0.44 0.43 0.56 0.31 − 0. 22 0.05 0.22 UD − 0. 15 0.05 − 0. 23 0.17 − 0. 15 − 0. 01 − 0. 02 − 0. 00 0.07 − 0. 02 − 0. 07 − 0. 21 − 0. 08 − 0. 20 − 0. 15 − 0. 12 − 0. 19 − 0. 13 − 0. 03 0.22 BOD 5 − 0. 24 − 0. 33 − 0. 34 0.19 0.00 0.12 − 0. 45 − 0. 47 − 0. 05 − 0. 11 0.18 − 0. 44 − 0. 11 − 0. 22 − 0. 27 − 0. 42 − 0. 23 − 0. 10 − 0. 23 0.29 (t year − 1) TSS (t year − 1) − 0. 24 − 0. 33 − 0. 34 0.19 0.00 0.12 − 0. 45 − 0. 47 − 0. 05 − 0. 11 0.18 − 0. 44 − 0. 11 − 0. 22 − 0. 27 − 0. 42 − 0. 23 − 0. 10 − 0. 23 0.29 Ntot (t year − 1) − 0. 24 − 0. 33 − 0. 34 0.19 0.00 0.12 − 0. 45 − 0. 47 − 0. 05 − 0. 11 0.18 − 0. 44 − 0. 11 − 0. 22 − 0. 27 − 0. 42 − 0. 23 − 0. 10 − 0. 23 0.29 Ptot (t year − 1) − 0. 24 − 0. 33 − 0. 34 0.19 0.00 0.12 − 0. 45 − 0. 47 − 0. 05 − 0. 11 0.18 − 0. 44 − 0. 11 − 0. 22 − 0. 27 − 0. 42 − 0. 23 − 0. 10 − 0. 23 0.29 Marked correlations are significant a t p < 0. 05 (N = 54 )

association between the obtained metric score and increased perturbation and vice versa.

From the original group of 184 metrics, a subset of 78 attributes met the criteria of low reference site within-site variability (low CVs). Inspection of zeros and low values in reference sites further refined this selection (“Electronic supplementary material”).

CCA analysis, including typological covariables following forward selection, resulted in 32 metrics that made a significant contribution ( p< 0.05) and were independent of natural variability. Ex-amination of VIFs of the successive CCAs analy-ses removed another 12 attributes.

The Pearson correlation coefficients for the 20 retained metrics (Table 5) showed that only % diap and % Cling showed no association with the selected disturbance variables (Table 5). Despite this, we retained all 20 metrics for further analysis, since they may have responded to variables that were not considered in this study. Examination of the boxplots revealed 12 the 20 retained candidate metrics to have little or no overlap between most disturbed and reference sites, satisfying the last metric selection criteria (Fig. 4). These included three structural metrics (fP, fEPT, and %5TD), seven adaptation metrics (fSwi, fCling, %exR, gShr, sFil, and fGath), and three metrics that

Fig. 4 Box plots of

macroinvertebrate metrics selected by CCA for reference and impaired sites, obtained separately for 1997 and 2000. Boxes are interquartile ranges (25%ile and 75%ile),

range bars show

maximum and minimum of nonoutliers, small

squares are medians, and dots are outliers

1997 2000 1997 2000 % Cling 0 20 40 60 80 100 %exR 0 20 40 60 80 100 fCling 0 3 6 9 12 15 18 fSwi 0 1 2 3 4 5 6 7 8 % diap 0 20 40 60 80 100 %br 0 20 40 60 80 100 %5TD 40 50 60 70 80 90 100 fEPT 0 5 10 15 20 25 fH 0 1 2 3 4 5 6 fP 0 1 2 3 4 5

Fig. 4 (continued) FBI 2 4 6 8 10 % aqP 0 20 40 60 80 100 %Int 0 20 40 60 80 100 %Hem 0 20 40 60 80 100 gShr 0 3 6 9 12 sFil 0 2 4 6 8 10 fGath 0 3 6 9 12 15 IBMWP 0 100 200 300 400 %CluV 0 20 40 60 80 100 %preR 0 20 40 60 80 100

Reference Impaired Reference Impaired Reference Impaired Reference Impaired

1997 2000 1997 2000

measure the tolerance level (%Int and two in-dices: IBMWP and FBI).

Discussion

To identify human impacts on aquatic assem-blages, measurements of the individual, com-munity coupled with landscape level scales are required (Butcher et al.2003) ensuring that when metrics are organized and selected systemati-cally within a regional framework, multimetric indices effectively measure changes along per-ceived disturbance gradients (Karr and Chu1999). This study included a large variety of point and nonpoint-source disturbances, including

agri-culture, pasture, deforestation and afforestation, physical alterations of riparian and instream habi-tats, road building, impoundments, urban areas, and mining. Ecological condition was described by a large array of metrics (functional characteristics, i.e., traits, and attributes derived from structural and condition aspects of communities) in order to identify which best assess impairment. In particu-lar, we wanted to test the possibility of integrating biological attributes into an assessment system.

In this study, it was verified an interannual vari-ability (Table2; Fig.3). These differences between years (1997 versus 2000) could be explained by the weather conditions over the analyzed period that corresponded to extreme events, such as droughts and floods. The year of 1997 was one of the fifth

hotter years of the last 150 years, and conversely, the winter of 2000 was the third rainiest of the last 30 years. Temporally, heterogeneous environ-ment is an event which communities are natu-rally dependent (Poof and Ward 1989; Hildrew and Giller 1994). According to Minshall (1988), changes in flow, light, temperature, resource sup-ply, and many other variables can occur over time scales, with varying degrees of predictability Poff (1992). However, in spite of the observed tem-poral changes in macroinvertebrate communities as a result of natural environmental variability, differences in spatial distribution of sampling sites across each year were not manifested for both years. Observing NMDS ordination and SIMPER analyses (Fig.3; Table2), the same environmental gradient structuring the characteristics of sites and the species occurring in them becomes clear. As a result, we can conclude that the dynamic nature of both environmental conditions and community structure in lotic environments do not have major effects on the sensitivity of the species and conse-quently in the sensitivity of this method.

Our results showed that 12 attributes success-fully distinguished between the reference and im-paired sites and also emphasized the value of combining conventional metrics (abundance and/ or richness of species, biotic indices, and tolerance measures) with biological traits (habit/behavior measures, habitat preference, and trophic groups) for reliable lotic biomonitoring. Fifty percent of the selected metrics were traits related to locomo-tion (2), preference of species in terms of current velocity (1), and functional feeding groups (3). Similar to other studies (Dolédec et al.1999; Poff 1997), our results confirm that, in the catchment context, species traits can be used to assess stress magnitude in running waters and have consider-able potential as a benchmark for biomonitoring and improving the traditional biomonitoring ap-proach. Rodgers et al. (1979) stressed that the integration of structural/compositional and func-tional metrics provides the best mean of assessing impairment. Phillips (2004) notes that, rather than simply record loss or reduction in numbers of species as a result of a disturbance, use of species traits allows us to identify the most sensitive life history characteristics. In an attempt to use the functional characteristics of lotic

macroinverte-brates for biomonitoring, Charvet et al. (1998) and Dolédec et al. (1999) found that the use of a large variety of biological traits was more reliable than community structure in detecting stream pol-lution in terms of species composition and abun-dances. Also, Charvet et al. (2000) concluded that community structure based on biological traits, unlike taxonomic composition, was relatively sta-ble across large environmental gradients (geology, altitude, geographical coordinates, stream order, and slope) enabling the use of these descriptors for biomonitoring over large geographic regions. According to Phillips (2004), another benefit of using complementary species traits is their stabil-ity through space and time, unlike species compo-sition. Vieira et al. (2006) refer as well that once some functional traits may not be constrained by taxonomy, they can be applicable at multiple spatial scales (local, regional, and continental) and can provide a consistent method for assessing community responses to environmental gradients. Thus, it becomes evident that assessment proto-cols based solely only on conventional attributes may not be sufficiently sensitive to distinguish im-pairment, whereas trait-based variables can detect problems that may otherwise be overlooked.

Over the last decade, the use of species traits has become more important in assessing stream system disturbance (Metcalfe1989; Townsend and Hildrew 1994; Dolédec et al. 1996,1999; Mabry et al. 2000; Ribera et al. 2001; Statzner et al.

2001a; Bady et al.2005; Dziock et al.2006).

How-ever, only a very small group of traits has been studied in combination with taxonomic indicators to produce macroinvertebrate based indices for biocriteria and assessment (Barbour et al. 1996; Fore et al. 1996; Kerans and Karr 1994). This paper has presented a heuristic framework that seeks, through a vast group of conventional met-rics and species traits, to choose the best combina-tions of attributes for multiscale and cumulative disturbance effects with a view to integrate this approach into future monitoring systems. There is, however, a lack of information for many en-demic invertebrate species concerning individual autecology (morphological and life history) and we support Poff’s (1997) comments that more research is needed to quantify species traits in stream organisms in order to obtain important

biological insights and better understanding of species-environmental relations. For this reason, future studies should focus on the documentation of clear relationships between biological traits and different human impacts on macroinvertebrates of running waters.

Our examination of all selected metrics has shown that “conventional”, well-tested, and used water quality assessment metrics such as the biotic indices IBMWP and FBI make a highly significant contribution to this work. These indices are based on the detection of organic pollution, based on a community’s response to high organic loading and decreased dissolved oxygen levels. Despite of some selected metrics have been developed solely for assessing organic pollution, many also can be used to define the tolerance of species to other human impacts. For example, species richness of Ephemeroptera, Plecoptera, and Trichoptera groups decreases not only with an increase of nu-trient concentration (Barbour et al.1996; Bratton et al.1980; Fitzpatrick et al. 2001) but also with an increase of contaminants, heavy metals, ther-mal pollution (Lillie et al. 2003), flow disruption (Fitzpatrick et al. 2001; Lillie et al. 2003), sed-iment input (Lenat 1984; Meban 2001; Quinn and Hickey 1990), acid rain (Peterson and Van Eeckhaute1992), and acid mine drainage (Zarger et al. 1986). Plecoptera (fP) are among the most sensitive indicator organisms and can indicate im-pairment resulting from low dissolved oxygen or siltation (Chirhart 2003). Measures of tolerance, such as %Int, indicate sensitivity of the assem-blage and component species to various types of perturbation (Hilsenhoff 1987). According to Chirhart (2003), the presence of moderate num-bers of intolerant taxa is an indicator of good aquatic health. Trophic and functional benthic structure measures (e.g., gShr, sFil, and fGath) are also useful since they are influenced by nu-trient enrichment (Fitzpatrick et al.2001; Quinn and Hickey 1990), disturbance of riparian corri-dors (Cummins 1973; Cummins et al.1989), and impoundment and regularization (Cortes et al.

2002a; Voelz and Ward 1991). Concerning

dom-inance measures and despite the difficulty of pro-viding guidelines for interpreting a community structure because of the existence of a myriad of natural communities (Lillie et al. 2003), results

have shown that the percentage of a dominant organism increases with increasing perturbation (Barbour et al. 1996; Chirhart 2003) notwith-standing they are not particularly sensitive to moderate amounts of organic or toxic loadings (Plafkin et al.1989). The percent contribution of dominant taxon metric (%5TD) proved to be a good indicator in our study area. Regarding habi-tat preference, Vieira et al. (2004) verified that rheophily and microhabitat preferences may be indicative of hydrological disturbance after a wild-fire. Rheophilous taxa has been also associated to small rivers without organic pollution (Usseglio-Polatera et al. 2000) confirming, this way, the ability of metrics like %exR in discriminating impairment.

It was difficult to establish a cause–effect relationship between traits that describe habi-tat/behavior (fSwi, fCling) and environmental variables due to the scarcity of information. How-ever, Charvet et al. (1998) mention that mobility via swimming may be sensitive to chemical con-tamination, and Chirhart (2003) highlights that a lack of clinger taxa can indicate siltation or substrate embeddedness as a result of erosion.

From all tested functional traits, only trophic-based and locomotion traits could distinguished between reference and disturbed sites. Previous studies from temperate zone frequently found that life-history-related traits (e.g., number of descen-dants per reproductive cycle, number of repro-ductive cycles/individual, life duration of adults) responded best to a range of anthropogenic stres-sors and those traits related to feeding strate-gies, body shape, and respiration generally were more weakly related to perturbations (Dolédec et al.1999,2006). A possible explanation for the absence of response in our study to life-history-related traits is that it is life-history-related to ecoregions. Portugal, inserted in the biogeographic region of Iberian Peninsula according to Illies (1978) and WFD (Annex XI), is a region with very different characteristics from those where the referred stud-ies took place. In spite of trait-based approaches rely on evolutionary responses to environmental selective forces across broad geographical gradi-ents (Dolédec and Statzner2008), Dolédec et al. (1999) refer that this type of traits can be con-strained by geographic latitude and alerts that

the future studies have to show over which ge-ographic range such a trait may serve in bio-monitoring. Moreover, our rivers are much less impacted than the studied large rivers in mid-Europe that have been largely altered by human activities. This could be another answer once the diverse expected communities has different life-history strategies and functional responses to ma-jor environmental characteristics that in our study did not reveal to be responsive to perturbation.

From the 12 retained metrics, 11 need to be identified to family or genus levels and only one needs the species-level resolution (sFil). This last one is a trait related to feeding strategies. In agreement with Dolédec et al. (1998, 2000) and Gayraud et al. (2003), trait-based approaches us-ing higher levels of taxonomic identification (for example, genus and family) may adequately de-scribe trait occurrence and may be more time efficient. Moulton et al. (2000) also refers that trait-based metrics also may be robust to tax-onomic ambiguities, which can influence how taxonomically based metrics respond to environ-mental gradients. Thus and according to these au-thors, we think that the best basis for assemblage description must be one of these two taxonomic levels or both, genus and family levels.

Consequent evaluations suggest that the se-lected metrics are good indicators of water qual-ity in the Douro catchment since they describe most of the variation verified in invertebrate assemblages and distinguish between different environmental conditions. The present set of met-rics has the advantage of integrating traits with “conventional” metrics, which makes them poten-tially able to assess multiscale effects, knowing that the human impacts acting at higher scales are successively modified (filtered) by local variables (Frissell et al. 1986; Poff 1997). Thus, this ap-proach provides a starting point for the selection of metrics that can be reliably applied to larger geographical areas beyond the river basin. Follow-ing screenFollow-ing and validation across a gradient of human influence, these metrics could be used in a refined index to better characterize community responses to general environmental degradation. Additionally, this information can be a baseline for developing strategies for the biomonitoring of aquatic ecosystem health and provide information

relevant to WFD, more specifically, with regard to its application in Portugal. Thus, the set of traits selected in this work could be used also to define assemblage types and determine expected biologi-cal conditions at reference sites for biomonitoring programs in a similar way to the ones that has been used at other European countries (Dolédec et al.1999,2000; Charvet et al.2000; Statzner et al.

2001b; Dziock 2006; Henle et al.2006; Dolédec

and Statzner2008).

Acknowledgment Samantha J. Hughes (Centre for

Macaronesian Studies, University of Madeira; Forest Re-search Centre, Technical University of Lisbon) provided comments on the first draft of this article and helped to improve the English.

References

Alba-Tercedor, J. (2000). BMWP’, un adattamento spag-nolo del British Biological Monitoring Working Party (BMWP) Store System. Biologia Ambientale, 14(2), 65–67.

Alba-Tercedor, J., & Sanchez-Ortega, J. A. (1988). Un método rápido y simple para evaluar la calidad bio-logica de las aguas Corrientes basado en de Hellawell, 1978. Limnetica, 4, 51–56.

Askew, R. R. (1988). The dragonflies of Europe (291 pp). Colchester: Harley Books.

Bady, P., Dolédec, S., Fesl, C., Gayraud, S., Bacchi, M., & Schöll, F. (2005). Use of invertebrate traits for bio-monitoring of European large rivers: The effects of sampling effort on genus richness and functional di-versity. Freshwater Biology, 50, 159–173. doi:10.1111/ j.1365-2427.2004.01287.x.

Barbour, M. T., & Yoder, C. O. (2000). The multimetric approach to bioassessment, as used in United States of America. In J. F. Wright, D. W. Sutcliffe, & M. T. Furse (Eds.), Assessing the biological quality of fresh

waters (pp. 281–292). Ambleside: The Freshwater

Biological Association.

Barbour, M. T., Stribling, J. B., & Karr, J. R. (1995). The multimetric approach for establishing biocriteria and measuring biological condition. In W. S. Davis & T. P. Simon (Eds.), Biological assessment and criteria: Tools

for water resource planning and decision-making

(pp. 63–77). London: Lewis.

Barbour, M. T., Gerritsen, J., Griffith, G. E., Frydenborg, R., McCarron, E., White, J. S., et al. (1996). A framework for biological criteria for Florida streams using benthic macroinvertebrates. Journal of the

North American Benthological Society, 15, 185–211.

doi:10.2307/1467948.

Barbour, M. T., Gerritsen, J., Snyder, B. D., & Stribling, J. B. (1999). Rapid bioassessment protocols for use

macroinvertebrates and fish (2nd ed.). EPA

841-B-99-002. Washington, DC: U.S. Environmental Protection Agency, Office of Water.

Bonada, N., Dolédec, S., & Statzner, B. (2007). Taxonomic and biological trait differences of stream macroin-vertebrate communities between Mediterranean and temperate regions: Implications for future scenarios.

Global Change Biology 13, 1658–1671.

Bonada, N., Prat, N., Resh, V. H., & Statzner, B. (2006). Development in aquatic insect biomonitoring: A com-parative analysis of recent approaches. Annual Review

of Entomology, 51, 495–523. doi:10.1146/annurev. ento.51.110104.151124.

Bratton, S. O., Mathews, R. C., & White, P. S. (1980). Agricultural area impacts within a natural area: Cade’s Cove, a case history. Environmental Management, 4, 433–448. doi:10.1007/BF01869654.

Butcher, J. T., Stewart, P. M., & Simon, T. P. (2003). A benthic community index for streams in the Northern lakes and forests ecoregion. Ecological

Indicators, 3, 181–193. doi:10.1016/S1470-160X(03) 00042-6.

Charvet, S., Kosmala, A., & Statzner, B. (1998). Biomon-itoring through biological traits of benthic macroin-vertebrates: Perspectives for a general tool in stream management. Archiv fuer Hydrobiologie, 142, 415–432.

Charvet, S., Statzner, B., Usseglio-Polatera, P., & Dumonts, B. (2000). Traits of benthic macroin-vertebrates in semi-natural French streams: An initial application to biomonitoring in Europe. Freshwater

Biology, 43, 277–296. doi:10.1046/j.1365-2427.2000. 00545.x.

Chinery, M. (1992). Insects d’Europe: Multiguide nature (380 pp). Paris: Bordas.

Chirhart, J. (2003). Development of a macroinvertebrate

in-dex of biological integrity (MIBI) for rivers and steams of the St. Croix River Basin in Minnesota (41 pp). St.

Paul, MN: Minnesota Pollution Control Agency, Bio-logical Monitoring Program.

Clarke, K. R., & Gorley, R. N. (2001). PRIMER v5: User

manual/tutorial. Plymouth: PRIMER-E.

Clarke, K. R., & Warwick, R. M. (2001). Change in

ma-rine communities: an approach to statistical analysis and interpretation (2nd ed.). Plymouth: Primer-e Ltd,

Plymouth Marine Laboratory.

Cortes, R. M. V. (1992). Seasonal pattern of benthic com-munities along the longitudinal axis of river systems and the influence of abiotic factors on the spatial struc-ture of those communities. Archiv fuer Hydrobiologie,

126, 85–103.

Cortes, R. M. V., Ferreira, M. T., Oliveira, S. V., & Godinho, F. (1998). Contrasting impact of small dams on the macroinvertebrates of two Iberian moun-tain rivers. Hydrobiologia, 389, 51–61. doi:10.1023/ A:1003599010415.

Cortes, R. M. V., Ferreira, M. T., Oliveira, S. V., & Oliveira, D. (2002a). Macroinvertebrate community structure in a regulated river segment with different flow conditions. River Research and Applications, 18, 367–382. doi:10.1002/rra.679.

Cortes, R. M. V., Oliveira, S. V., Cabral, D. A., Santos, S. & Ferreira, T. (2002b). Different scales of analysis in classifying streams: From a multimetric towards an integrate system approach. Large Rivers 13. Archiv für

Hydrobiologie Suppl., 141(3–4), 209–224.

Cummins, K. W. (1973). Trophic relations of aquatic in-sects. Annual Review of Entomology, 18, 183–206. doi:10.1146/annurev.en.18.010173.001151.

Cummins, K. W., Wilzbach, M. A., Gates, D. M., Perry, J. B., & Taliaferro, W. B. (1989). Shredders and riparian vegetation. Bioscience, 39, 24–30. doi:10.2307/ 1310804.

Davis, W. S., Snyder, B. D., Stribling, J. B., & Stoughton, C. (1996). Summary of state biological assessment

programs for streams and rivers. EPA 230-R-96-007.

Washington, DC: Office of Planning, Policy, and Eval-uation, US Environmental Protection Agency. De Pauw, N., & Vanhooren, G. (1983). Method for

biologi-cal quality assessment of watercourses in Belgium.

Hy-drobiologia, 100, 153–168. doi:10.1007/BF00027428. Dolédec, S., Chessel, D., & Ter Braak, C. J. F. (1996).

Matching species traits to environmental variables.

Environmental and Ecological Statistics, 3, 143–166.

doi:10.1007/BF02427859.

Dolédec, S., Olivier, J. M., & Statzner, B. (2000). Accurate description of the abundance of taxa and their biolog-ical traits in stream invertebrate communities—effects of taxonomic and spatial resolution. Archiv fuer

Hydrobiologie, 148, 25–43.

Dolédec, S., Philips, N., Scarsbrook, M., Riley, R. H., & Towsend, C. R. (2006). Comparison of structural and functional approaches to determining landuse ef-fects on grassland stream invertebrate communities.

Journal of the North American Benthological Society, 25(1): 44-60. Cote bibliothèque 06.3.

Dolédec, S., & Statzner, B. (2008). Invertebrate traits for the biomonitoring of large European rivers: An as-sessment of specific types of human impact.

Fresh-water Biology, 53, 617–634. doi:10.1111/j.1365-2427. 2007.01924.x.

Dolédec, S., Statzner, B., & Bournard, M. (1999). Species traits for future biomonitoring across ecore-gions: Patterns along a human-impacted river.

Fresh-water Biology, 42, 737–758. doi:10.1046/j.1365-2427. 1999.00509.x.

Dolédec, S., Statzner, B., & Frainay, V. (1998). Accu-rate description of functional community structure: Identifying stream invertebrates to species-level?

Bul-letin of the North American Benthological Society, 15,

154–155.

Dziock, F. (2006). Life-history data in bioindication pro-cedures, using the example of Hoverflies (Diptera, Syrphidae) in the Elbe floodplain. International

Re-view of Hydrobiology, 91, 341–363. doi:10.1002/iroh. 200510889.

Dziock, F., Henle, K., Foeckler, F., Follner, K., & Scholz, M. (2006). Biological indicator systems in floodplain— a review. International Review of Hydrobiology, 91, 271–291. doi:10.1002/iroh.200510885.

European Commission (2000). The European Water Framework Directive 2000/60/CE (WFD) of 23

October 2000 establishing a framework for a commu-nity policy in the domain of water. Official Journal of

European Communities, L327, 1–72.

Faessel, B. (1985). Les Trichoptères. Données biologiques, éthologiques et écologiques. Clés de détermination larvaire des familles et des principaux genres de France. Bulletin Francais de la Peche et de la

Piscicul-ture, 299, 1–41. doi:10.1051/kmae:1985001.

FAME Consortium (2004). Manual for the application of

the European Fish Index—EFI. A fish-based method to assess the ecological status of European rivers in support of the Water Framework Directive. Version

1.1.http://fame.boku.ac.at.

Fitter, R., & Manuel, R. (1994). Lakes, rivers, streams

and ponds of Britain and North-West Europe: Collins Photo Guide (382 pp). London: Harper Collins.

Fitzpatrick, F. A., Scudder, B. C., Lenz, B. N., & Sullivan, D. J. (2001). Effects of multi-scale environmen-tal characteristics on agricultural stream biota in eastern Wisconsin. Journal of the American Water

Resources Association, 37, 1489–1507. doi:10.1111/j. 1752-1688.2001.tb03655.x.

Fore, L. S., Karr, J. R., & Wisseman, R. W. (1996). Assessing invertebrate responses to human activi-ties: Evaluating alternative approaches. Journal of the

North American Benthological Society, 15, 212–231.

doi:10.2307/1467949.

Frissell, C. A., Liss, W. J., Warren, C. E., & Hurley, M. D. (1986). A hierarchical framework for stream habitat classification: Viewing streams in the water-shed concept. Environmental Management, 10, 199– 214. doi:10.1007/BF01867358.

Gayraud, S., Statzner, B., Bady, P., Haybachp, A., Sholl, F., Usseglio-Polatera, P., et al. (2003). Inverte-brate traits for the biomonitoring of large European rivers—an initial assessment of alternative metrics.

Freshwater Biology, 48, 2045–2064. doi: 10.1046/j.1365-2427.2003.01139.x.

Henle, K., Dziock, F., Foeckler, F., Follner, K., Hüsing, V., Hettrich, A., et al. (2006). Study design for as-sessing species environment relationships and de-veloping indicator systems for ecological changes in floodplains—the approach of the RIVA Project.

International Review of Hydrobiology, 91, 292–313.

doi:10.1002/iroh.200610886.

Hildrew, A. G., & Giller, P. S. (1994). Patchiness, species interactions and disturbance in the stream benthos. In P. S. Giller, A. G. Hildrew, & D. G. Raffaelli (Eds.),

Aquatic ecology: Scale, pattern and process (pp. 21–

62). London: Blackwell Scientific.

Hill, M. O. (1973). Diversity and evenness: A unifying notation and its consequences. Ecology, 54, 427–432. doi:10.2307/1934352.

Hill, M. O. (1979). DECORANA-A FORTRAN program

for detrended correspondence analyses and reciprocal averaging (p. 95). Ithaca, NYY: Cornell University.

Hill, M. O., & Gauch, H. G. (1980). Detrended correspon-dence analysis, an improved ordination technique.

Vegetatio, 42, 47–48. doi:10.1007/BF00048870. Hilsenhoff, W. L. (1987). An improved index of organic

stream pollution. Great Lakes Entomologist, 20, 31–39.

Hilsenhoff, W. L. (1988). Rapid field assessment of organic pollution with a family-level biotic index. Journal of

the North American Benthological Society, 7, 65–68.

doi:10.2307/1467832.

Hynes, H. B. N. (1979). The ecology of running waters (3a

ed., p. 555). Liverpool: Liverpool University Press. Illies, J. (Ed.) (1978). Limnofauna Europaea (2nd ed.).

Stuttgart: G. Fischer.

Illies, J., & Botosaneanu, L. (1963). Problèmes et méth-odes de la classification et de la zonation écologique des eaux courantes, considérées surtout du point de vue faunistique. Verhandlungen - Internationale

Vere-inigung für Theoretische und Angewandte Limnologie, 12, 1–57.

Karr, J. R. (1991). Biological integrity: A long neglected aspect of water resource management. Ecological

Applications, 1, 66–84. doi:10.2307/1941848.

Karr, J. R., & Chu, E. W. (1999). Restoring life in

run-ning waters: Better biological monitoring. Covelo, CA:

Island.

Karr, J. R., Fausch, K. D., Angermeier, P. L., Yant, P. R., & Schlosser, I. J. (1986). Assessing biological integrity

in running waters: A method and its rationale.

Spe-cial publication 5. Urbana: Illinois Natural History Survey.

Kashian, D. R., & Burton, T. M. (2000). A compari-son of macroinvertebrates of 2 Great Lakes Coastal Wetlands: Testing potential metrics for an index of ecological integrity. Journal of Great Lakes Research,

26, 460–481.

Kerans, B. L., & Karr, J. R. (1994). A benthic index of biotic integrity (B-IBI) for rivers of the Tennessee Valley. Ecological Applications, 4, 768–785. doi:

10.2307/1942007.

Lenat, D. R. (1984). Agriculture and stream water qual-ity: A biological evaluation of erosion control pro-cedures. Environmental Management, 8, 333–343. doi:10.1007/BF01868032.

Lenat, D. R. (1993). A biotic index for the southeastern United States: Derivation and list of tolerance values, with criteria for assigning water-quality ratings.

Jour-nal of the North American Benthological Society, 12,

279–290. doi:10.2307/1467463.

Lillie, R. A., Szczytko, S. W., & Miller, M. A. (2003).

Macroinvertebrate data interpretation guidance man-ual. Madison: Wisconsin Department of Natural

Resources.

Mabry, C., Ackerly, D., & Gerhardt, F. (2000). Land-scape and species-level distribution of morphologi-cal and life history traits in a temperate woodland flora. Journal of Vegetation Science, 11, 213–224. doi:10.2307/3236801.

Margalef, R. (1951). Diversidad de especies en las comu-nidades naturales. Publicaciones del Instituto de

Biolo-gia Aplicada, Barcelona, 6, 59–72.

Margalef, R. (1983). Limnologia. Barcelona: Omega. Meban, C. A. (2001). Testing bioassessment metrics:

Macroinvertebrate, sculpin, and salmonid responses to stream habitat, sediment, and metals.

Environ-mental Monitoring and Assessment, 67, 293–322.